Re-evaluating Mpeg Motion Compensation Search Criteria

As processors evolve from simple scalar machines to more advanced designs using dynamic instruction scheduling, speed critical algorithms should be reexamined to ensure that the optimal strategy is being used. One important example of this class of algorithms is MPEG encoding. Although the Mean Absolute Distance approach is recognized in the literature as more eecient in terms of speed, our research indicates that the Mean Square Error approach becomes much more attractive as instruction level parallelism increases. In a machine that can exploit a high degree of instruction level parallelism, the Mean Square Error approach yields better performance. In addition, the Mean Square Error approach achieves better compression quality than the Mean Absolute Distance method.